Graduate Student Research Assistants - Responsibilities and Best Practices

Overview

The opportunity to work as a Graduate Student Research Assistant (GSRA) is one of many important strengths of our department. The department can provide this opportunity because our faculty have been successful at obtaining grant funding as principal investigators and in convincing research collaborators that co-mentoring our Biostatistics graduate students is rewarding and worth the substantial cost of stipend and tuition owing to the knowledge, creativity, and work ethic of our students. Continued GSRA support for present and future students requires that students perform well as GSRAs, so that department faculty and our collaborators continue to obtain and renew these positions. It is equally important for GSRAs to perform well in coursework:in fact, satisfactory progress towards degree requirements and a grade average of better than B+ are required for a student to remain eligible for a GSRA position. 

This document is intended to help students and their faculty advisors navigate the dual roles of graduate student and GSRA by providing guidance for best GSRA practices. However, please note that each GSRA position comes with its own requirements and expectations. Early and frequent communication between the supervisor and the GSRA is vital to achieve a productive working relationship.

Hours of Employment

GSRA appointments within the Department are at 50% effort, requiring 20 hours of work per week. Some GSRA positions require this 20 hours per week be consistent across the semester. More commonly, hours are more flexible. Somewhat more hours might be expected by the supervisor during key weeks with abstract and grant deadlines, while lighter hours might be requested by the student during exam weeks or other busy times. GSRAs and their faculty supervisor should discuss this issue at the beginning of their time working together. Communicating about deadlines/timelines of importance to both the supervisor and the GSRA can help with managing priorities and keeping work expectations both reasonable and productive. 

Spring and summer months (May to August) are typically the most productive time for a GSRA while also being less costly for the supervisor, as no tuition is charged in that time period. Therefore, even though students are not guaranteed funding in the summer, GSRAs are generally expected to work throughout the year.  GSRAs are guaranteed a minimum of 10 business days of vacation per year; always check with your supervisor(s) before scheduling vacations/time off. Many supervisors are flexible about time away, particularly if work meetings for pressing deadlines can be held via conference call or videoconferencing during time away. Summer internships or other extended absences that interrupt GSRA work typically are not allowed, although this is again something that can be discussed with your GSRA supervisor. Expectations about vacations and summer internships are again topics that should be discussed during the first GSRA/supervisor meeting so that you and your supervisor can  plan accordingly.   

General Expectations

GSRAs are expected to conduct themselves professionally as a member of the research team and to proactively work to advance the research project(s). This includes working to understand the broad goals and activities of the project and their own specific task(s). The GSRA is not expected to succeed at their task without help, but they are expected to identify barriers to progress and proactively seek help to address those barriers.

Onboarding

Becoming a useful part of the research team as quickly as possible is important. Upon learning about the GSRA appointment, the student should schedule a meeting with the primary supervisor to learn about the upcoming work and discuss the structure of the working relationships. Here is a (non-exhaustive) list of strategies that can help you to get up to speed as fast as possible:

  • Your goal should be to try some analyses early on and present some preliminary results as soon as possible. These initial results will typically not be perfect but instead be the start of an iterative process that will lead to final results and help you understand key aspects of the project.  
  • Information overload is typical during the first stage of a GSRA appointment. Prioritize information that allows you to get going with your work; many details will become clearer once you get started on your data analysis or methods problem. Ask for help to prioritize your learning.
  • Make sure you understand the computer environment used for the project. You may need to download new software or even familiarize yourself with a new programming language. Identify resources that help you get started.    
  • Ask about available resources for understanding your project. Often, there will be a set of papers that provides the foundation of your project. You should continue to deepen your mastery of the biological/medical aspects and the statistical aspects of the project throughout your employment. 
  • Current or previous GSRAs working for the same supervisor should be willing to help with advice, code, and encouragement. These GSRAs have useful experience with the study team, may have worked with some of the same datasets, and may have experience with standard or non-standard analysis methods favored by the supervisor. 
  • If the research topic or an analysis method is new to you, ask for reference materials to get up to speed. Ask if a previous GSRA has worked with the topic/dataset/method before and might be willing to mentor during the early stages of a project.
  • Some supervisors maintain an MBox with SAS, R, or other relevant code written by GSRAs for previous projects and/or reading material that can be helpful in getting oriented to the job. Soon, you should be contributing well commented code and examples of how to do new analysis techniques to the repository of information available to current and future GSRAs, staff, and faculty.
  • Following a meeting while you are new to a project can be challenging. Ask if it is okay to record meetings so that you can review your notes at a later time. Take snapshots of handwritten notes for reference and for sharing with the group, as appropriate. 
  • If you are working on a project with an external collaborator, it may be useful to establish some initial protocols for communication between you, your advisor, and the collaborator. In particular, you want to achieve clarity whether collaborators typically contact you directly with requests, or whether such requests go through your advisor. Similarly, you may want to establish whether you should communicate results directly to the collaborator or whether they should be assessed by your advisor first.

Meetings

Regular meetings constitute a necessary time investment by the GSRA and supervisor and their collaborators, if any. If necessary, set up a regularly scheduled meeting that fits with everyone’s schedule and send an invitation to other team member calendars (if this is not done by someone else on the team). Here are some steps to maximize the utility of this time investment.

  • When describing your work, it is often useful to briefly indicate about how much time you invested. This allows your advisor to make sure you are not overburdened and may also help them identify inefficiencies with your approach.
  • Prepare well for each meeting, be on time, listen carefully, and ask questions. During meetings, you will likely be asked to report on your progress since the last meeting. Bring your laptop to each meeting and be prepared to share your screen with the research team. Slides are often a good means of describing the work done.
  • The art of clearly communicating statistical ideas to a non-statistically minded collaborator can be challenging and is an important skill to develop. Don’t be frustrated when you are having a hard time making yourself understood. Take note of the approaches used when individuals with different skill sets communicate successfully, for example, when a mentor teaches a collaborator how to interpret results of analyses or explains why one analysis is preferred to another.
  • During meetings, you will often be given new tasks. Try to make sure you understand the requested tasks. If specifics of the task are unclear, ask during the meeting or follow-up afterwards to get the necessary information. During brainstorming meetings, many potential ideas may be suggested. Be sure to review action items before the meeting ends and clarify ‘nice to have’ and ‘must have’ tasks and their deadlines.
  • If a regular meeting time cannot be found, always try to set the time for the next meeting before the meeting ends. Meetings can occasionally be rescheduled if the GSRA or supervisor is unavoidably not prepared, but this should not occur often.
  • Even if the current project appears finished, consider setting a time for brainstorming about the next project to help keep things moving forward with the research team.

Data Handling

Many GSRA positions will involve the analysis of biomedical data. Such data are reliably less organized and more complex than expected, often creating complications during the analysis. Strategies to minimize these complications include:

  • Never assume the dataset provided is without errors. Check for missing values, unrealistic data values, inconsistencies in dates, etc. When you come across data values with apparent problems, note the subject IDs that have issues. Often the person who provided the data can go to the original data source for clarification and correction of data quality issues.
  • If merging data from multiple sources to produce the analysis data set, keep all original datasets together in the same place. Save a copy of the merged dataset along with code that produced it.
  • Appropriately guard confidential data. THIS IS EXTREMELY IMPORTANT. Understand ownership and restrictions of the data you are analyzing. Do not share the data with others without the explicit permission of the owners. Do not submit abstracts or make presentations based on the data without making sure all data owners have seen and agreed to your abstract or presentation materials. Make sure your accounts containing such data are secured with strong passwords.

Programming/Computational Analysis

Much of your GSRA work will be based on computer programs created by you or others. Efficiency can be gained for you and your team members by following some basic concepts:

  • Organize and comment your code with future programmers in mind. It is the GSRA’s responsibility to produce and pass on a well commented version of code/data/analyses used in submitted/published manuscripts. Use interpretable variable names, for instance, ‘female’ is better than ‘gender’. Indicate units clearly, for instance, ‘height_ft’ is better than height.
  • If possible, organize code in a format that allows for easy editing. Thus, if your supervisor requests additional tweaks during a meeting you may be able to run the code from your laptop prior to the meeting. For instance, GSRAs should be able to run additional regression models by copying and pasting an existing model and editing in new variable names the supervisor wishes to explore. GSRAs in early stages of research are not typically expected to program ‘live’ in meetings, but with further experience this skill is extremely valuable in moving projects forward to publication. Coding alongside your mentor in live meetings can be a terrific opportunity to grow as a programmer and research collaborator. Always verify code that was programmed ‘live’ in meetings to make sure any data quality or coding errors are caught prior to moving forward with the project.
  • If your work makes use of a sequence of existing programs, create a script that runs this sequence. This will make it faster for you to re-run an analysis and by saving the script you will more easily be able to remember how the analysis was done.
  • As the project progresses, archive old code and create new scripts that correspond to the most current version of the project/manuscript draft. At the time of either manuscript or revision submission, the dataset used to produce tables, figures, etc. cited in the manuscript should be frozen with no further updates made so that results cited in the manuscript can be reproduced at any time. Code should be organized and commented according to the order the results are given in the final version of the paper. For instance, ‘code for Table 1’, followed by the corresponding R or SAS code. Well-commented coding will make it easier for future GSRAs to orient themselves to the research team and verify their own code is working correctly.
  • When using computational resources such as the Biostatistics or Center for Statistical Genetics cluster, be respectful of other users. Make sure that you understand how to submit jobs and learn how to use computational resources effectively. Consider which results you save to disk, as filling up the cluster’s hard disk will make the cluster inaccessible for everyone. Before running a computationally challenging job on e.g. the whole genome, analyze only a subset (e.g. chromosome 20) to make sure your scripts work as expected. If possible, test new software on a desktop before running it on a cluster. 

Scientific Communication in Writing

An important aspect of research is the ability to disseminate research results to a wider audience. As a GSRA, you may be asked to draft (segments of) a paper that describe your work and to create publication-ready figures of your results. For both non-native and native speakers, improving English communication skills has short- and long-term benefits that are worth the additional effort that may be required. International students should take advantage of opportunities to speak with and listen to English when possible, and ask other international students on campus for advice/resources for improving English language skills.

For all GRSAs, here is a list of approaches on how to work on your writing skills:

  • Read through previously published papers by the group for examples of good writing, especially well-written  statistical Methods and Results sections.
  • Be proactive in producing sentences, paragraphs, or sections that might be included in early manuscript drafts. Leverage the resources provided by the U of M for improving your writing, including the SPH Writing Lab and Sweetland Writing Center. Be prepared to taking on the drafting of a paper should the opportunity arise.
    Creating publishable figures often requires plotting the same data in multiple different ways to identify the display that best supports the claims of the publication. During the early stages of a project, it is typically acceptable to show drafts of tables and figures as opposed to highly polished tables and figures. However, as the final figures and analyses are identified for a manuscript, it is important for the GSRA to produce publication quality tables and figures. After you have settled on the type of display item you are creating consider the following aspects:
    • Tables and figures should not include variable names used in code, but the appropriate label. For instance, ‘height_ft’ should be replaced with ‘Height (feet)’.
    • Always make sure that units are displayed and that reference groups are clearly identified (as appropriate). It may be necessary to change units to a more interpretable scale when reporting parameter estimates (years typically better than days, for instance).
    • Make sure that font sizes used in tables and figures are large enough to be legible when reduced in size for a publication.
    • Double-check whether figures will appear in color or in black and white in the published manuscript. Legends and line types used in figures should be interpretable if converted from color to black and white.
    • Cross-reference numbers cited in the main text with numbers shown in tables and figures to make sure these match one another and are based on the most up-to-date version of analyses.
    • Look for obvious typos/inconsistencies. Do numbers displayed in tables add up correctly to the total number of subjects in the manuscript?  Is a consort diagram needed to describe why sample sizes are different between analyses?  Are displayed confidence intervals consistent with the significance of p-values shown in tables or cited within the manuscript?

Do not regard this list as a completely thorough reference for writing. Please consult your advisor for additional advice. 

Reporting a Problem With Your GSRA Position

Although most GSRAs in our department have positive experiences working with and learning from their faculty supervisor(s), issues sometimes do arise. Examples include: overwork, unclear expectations for work or time off, dissatisfaction with work topic, personality or other sort of mismatch with supervisor, and discrimination, harassment, or emotional/verbal abuse. When these problems occur, if possible GSRAs should attempt to resolve them by speaking with their supervisor directly as a first step. Problems can often be resolved through an open conversation. However, if the direct approach fails, or if the problem is so severe that one-on-one negotiation is impossible (e.g. harassment or abuse), GSRAs have several options to try and improve their situation.

Students can meet with a faculty or staff member they trust to try and resolve the situation. If they do not feel comfortable doing so, they can also use UM conflict resolution resources, such as the Office of Student Conflict Resolution (OSCR), Graduate Employee’s Organization (GEO), and the Office of Institutional Equality (OIE). Please note that GEO does not directly support GSRA grievances, but they will provide resources and advocate for you in the department. Students may also request a meeting with the Biostatistics Student Conflict Resolution Committee (BSCRC) to discuss their problem and future steps.

As in any workplace, you should expect your supervisor to treat you with respect and courtesy, and you should do the same to them. Although you may do so if you wish, you are not expected to communicate with your supervisor by text, frequently attend weekend or evening meetings, or comply with last-minute requests issued during non-business hours. For more information on the guidelines for graduate students to report discrimination and/or harassment, please refer to the website from Rackham Graduate School. Please note that if you report sexual harassment through the form given above, the recipient of the report is required by University of Michigan policy to report the claim to the Title IX Office, so that for reports of sexual harassment, confidentiality WILL NOT be guaranteed. 

Biostatistics Student Conflict Resolution Committee

The Biostatistics Student Conflict Resolution Committee (BSCRC) helps students resolve issues with their advisor, faculty member, or fellow student. Committee members serve as confidential, judgement-free peer mentors for students. Students may reach out to the committee if they need help voicing a concern, advocating for their rights, or expressing dissatisfaction. We prioritize the health and well-being of the students, and will strive to find the best solution for all involved. The committee also provides information on University conflict resolution resources available to students. Students can email biostat.crc@umich.edu or Nicky Wakim (nwakim@umich.edu) to set up appointments and request more information. Emails sent to the committee will only be received by student members. No faculty/staff members will receive emails.